Ketamine Treatment and Global Brain Connectivity in Major Depression

Resting-state functional connectivity MRI (rs-fcMRI) has revealed widespread alterations of large-scale brain networks in major depressive disorder (MDD), but most prior work used seed-based approaches that require preselecting regions of interest. Abdallah and colleagues adopted a fully data-driven, graph-based metric—global brain connectivity with global signal regression (GBCr)—to quantify whole-brain functional dysconnectivity without a priori seeds. GBCr indexes the average correlation of each voxel's BOLD time series with all other grey-matter voxels and has been related to regional cerebral blood flow and normal cognitive function in earlier studies. This study set out to compare GBCr in medication-free patients with MDD during a depressive episode versus healthy controls, and to test whether a single subanaesthetic infusion of ketamine (0.5 mg/kg) would rapidly normalise GBCr abnormalities. The investigators hypothesised that MDD patients would show reduced prefrontal cortex (PFC) GBCr at baseline and that ketamine's rapid antidepressant effects—typically evident within 24 hours—would parallel increases in PFC GBCr, with greater GBCr change in clinical responders.

Participants were male and female adults aged 21–65 years. Eighteen patients meeting DSM-IV criteria for current MDD with at least moderate severity (Inventory of Depressive Symptomatology-Clinician Rated score ⩾ 32) and documented nonresponse to at least two adequate antidepressant trials were enrolled, together with 25 healthy control (HC) volunteers free of lifetime psychiatric illness. Exclusion criteria included current alcohol or substance abuse disorder and lifetime bipolar or psychotic disorder. Patients were medication-free for at least one week prior to imaging; benzodiazepines were permitted as needed but withheld on scan days. Urine drug screen, medical stability and MRI safety were required. After baseline clinical assessment and structural and resting-state MRI, MDD participants received a single intravenous infusion of ketamine hydrochloride 0.5 mg/kg over 40 minutes in a monitored clinical research setting. Resting-state BOLD scans were repeated 24 hours after infusion. Depression severity was rated at baseline and 24 hours by independent, blinded raters using the Montgomery–Åsberg Depression Rating Scale (MADRS); response was defined as ⩾ 50% reduction in MADRS score. Imaging used a Philips 3.0 T scanner with standard high-resolution T1-weighted anatomical scans and a resting-state BOLD acquisition (TR = 2000 ms; 120 frames; voxel dimensions ≈2.2 × 2.2 × 3.25 mm). Participants rested with eyes open. GBCr for each voxel was calculated as the average correlation between that voxel's BOLD time series and all other grey-matter voxels; the authors note prior methodological details and rationale in supplementary material. Statistical analyses compared demographic features with t-tests and chi-square tests. Voxel-wise GBCr contrasts used independent t-tests for between-group comparisons and paired t-tests for within-subject pre/post treatment comparisons. Delta GBCr (post minus pre) was compared between responders and non-responders. Type I error was controlled by peak and cluster-extent correction; all tests were two-tailed with significance at p ⩽ 0.05. Non-parametric permutation testing (5,000 permutations) was used to confirm responder-related findings.

Eighteen MDD patients and 25 healthy controls completed the protocol. Groups did not differ significantly in age, gender, race or education. The MDD sample had a chronic, treatment-refractory illness course. Ten of 18 patients (56%) met the response criterion at 24 hours post-ketamine. Depression severity decreased significantly after treatment: mean delta MADRS = 14.2 ± 1.9, t(17) = 7.6, p = 0.000001. At baseline, voxel-wise whole-brain comparisons revealed widespread GBCr abnormalities in MDD relative to HC. Seven PFC clusters showed significantly reduced GBCr in MDD. Areas of relatively increased GBCr in MDD at baseline included the posterior cingulate, precuneus, lingual gyrus/occipital cortex and extensive cerebellar regions. Across the brain, 2,174 voxels showed altered GBCr in the preketamine MDD vs HC contrast (corrected p < 0.05). Twenty-four hours after ketamine, GBCr abnormalities were markedly reduced in spatial extent: only 310 voxels differed between MDD and HC, of which 123 overlapped with the baseline set. Paired within-subject tests showed significant post-treatment increases in GBCr in the right lateral PFC and reductions in the left cerebellum. Changes in GBCr related to clinical response. Delta GBCr was higher in responders than non-responders in the right lateral PFC and the left anterior insula; a 5,000-permutation non-parametric test produced similar clusters. No baseline GBCr differences distinguished future responders from non-responders. Postketamine, responders exhibited significantly higher GBCr than non-responders in five clusters: bilateral caudate, bilateral lateral PFC and left middle temporal cortex, and individual-level plots related delta GBCr in these clusters to percentage MADRS improvement. Distribution analyses of PFC GBCr showed a large effect-size leftward shift (reduced PFC GBCr) in preketamine MDD vs HC: Cohen's d = 0.95 (CI 0.91–0.99). Postketamine, the PFC GBCr distribution in responders overlapped more with controls (d = 0.46, CI 0.43–0.50), whereas non-responders retained a large left shift (d = 0.92, CI 0.87–0.97). Follow-up seed-based analyses provided preliminary circuit-level detail. Using a priori seeds in sgACC, DLPFC and PCC, MDD subjects demonstrated increased short-range connectivity within the PFC but reduced long-range connectivity between PFC and the rest of the brain; ketamine reduced within-PFC connectivity and enhanced PFC-to-other-region connectivity. Data-dependent seeds in lateral PFC and caudate (clusters that increased GBCr in responders) showed that, compared with non-responders, responders had greater postketamine lPFC and caudate connectivity with regions outside the PFC/subcortex but lower within-PFC/subcortical connectivity.

Abdallah and colleagues interpret the findings as evidence that MDD during a depressive episode is characterised by reduced prefrontal global connectivity and that a single ketamine infusion largely normalises these PFC GBCr abnormalities within 24 hours. Responders showed more pronounced increases in PFC, caudate and insula GBCr, implicating these frontal and striatal circuits in successful antidepressant response. Elevated baseline GBCr in posterior midline and occipital regions and extensive cerebellar hyperconnectivity were noted and showed partial or full normalisation after treatment. The authors discuss two non-mutually exclusive interpretations. One is that GBCr abnormalities are state-dependent and reverse with successful treatment regardless of treatment modality; the alternative is that GBCr partly reflects trait-related pathology that ketamine specifically targets. They propose, drawing on preclinical studies, that GBCr may under some circumstances index regional synaptic strength because of its positive association with regional cerebral blood flow and metabolic measures; thus reduced PFC GBCr could reflect chronic stress-related synaptic loss, and rapid restoration of GBCr after ketamine may mirror synaptogenic effects observed in animal models. The paper situates the results within broader literature showing PFC GBCr reductions across several chronic stress–related psychiatric disorders and notes differences in the anatomical distribution of abnormalities across conditions. The authors also propose a mechanistic model in which depression is associated with increased local (short-range) connectivity within PFC/subcortex and reduced long-range connectivity with the rest of the brain, and ketamine restores long-range integration and the PFC's central role in global network function. Key limitations are acknowledged: the sample included comorbid anxiety disorders which might have affected GBCr; the medication-free period was brief so residual effects of prior antidepressants cannot be excluded; resting-state acquisitions were relatively short; and the responder subgroup analysis had limited sample size making those findings preliminary. The authors further note that GBC without global signal regression did not detect MDD abnormalities in this cohort, and that the study examined only a single post-treatment time point so duration of connectivity changes is unknown. They call for multimodal and larger studies, including placebo and conventional antidepressant comparators, to test specificity, to relate GBCr to synaptic measures (for example using SV2A PET or metabolic spectroscopy), and to determine the temporal persistence and clinical utility of GBCr as a biomarker for rapid-acting antidepressants.